You are right, IF the early markets are still not profitable - the key numbers that we don't have are: how are the first 100 cities they launched doing? how are the first 10 cities they launched doing? If those cities have achieved profitability or are approaching it, then it works. Unfortunately, we don't have those #s.
However, if you look here there's some indication that earlier markets were at least closing the operating loss gap:
QQ4.12: 8.2M / 7.0M Q1.13: 12.9M / 7.3M Q2.13: 19.3M / 8.1M
Losses slowed dramatically here, while revenue grew. Note, Uber was still launching cities at this point, but had already taken most of the major US markets - my guess is that SF + NY + LA were getting closer to paying for themselves while they streamlined their launch process into smaller markets in the US. I feel OK making that guess because it's hard to believe that an investor would not have asked for that, and it's hard to believe that an investor would value Uber at $3.5B (the valuation at that time) if that wasn't happening. (Note - I understand that investors are not always actually this rational, I'm just making an assumption in this case).
I imagine the pitch was: look at our early markets, they are cash-cows, and we know how to make them. Now, give us $XB to go make more of these cash-cows please and don't ask us about losses for a long time, thanks! Which is why they don't really care about these leaked #s.